A Foundational Look
It’s not AI. It’s unexamined trust.
AI is inside your business right now, invited or not, with a company policy or without one. The misuse of the tool is the greatest risk.
Four patterns, all of them growing
Blind trust
A language model is built to be fluent, confident, and agreeable, not to be right. It will state a wrong answer with the same polish as a correct one, and it is tuned to tell its user what they want to hear. “The AI said so” is now standing in for judgment in businesses that, until recently, thought for themselves.
No guardrails
Employees are pasting client lists, financials, and contract terms into public AI tools because no one ever said not to. There is no usage policy, no data rule, no answer to the question “who is accountable for this output?”
The AI‑armed negotiation
Compensation and ownership conversations are arriving pre‑loaded: polished, machine‑built cases that sound authoritative and know nothing about your business, your margins, your values, or your history with the person across the table. AI can be a tool that makes the user more certain and less informed. That is a dangerous combination in a negotiation.
Machines talking to machines
One employee has AI draft the email. The recipient has AI summarize it and draft the reply. The document moves through the company and the issue gets closed, but the problem persists. The volume of work goes up. Human oversight diminishes.
The damage is subtle, but it compounds
None of this shows up on a P&L at first. What degrades is the thing a small business actually runs on: the quality of its decisions. Wrong numbers get repeated until they become the plan. Confidential information leaves the building without a trace. People stop checking, because the output appears complete.
Big companies can absorb a bad decision. Small businesses simply cannot. Judgment is the safeguard, and right now, in company after company, it is being outsourced by default, one convenient cut-and-paste at a time.
AI adoption, in practice
I am not anti‑AI, and I am not selling you an AI transformation. Used with discipline, these tools can be powerful time savers. But discipline is exactly what’s missing, and it doesn’t come from a subscription. It comes from the same place every other system in your business comes from:
Understanding the instrument. Your team needs to know, in plain terms, what a language model is and is not, so they stop treating a pattern machine like a knowing machine.
Rules for data. Every company needs a clear policy on what may leave the building, what may not, and through which doors.
Accountability and verification. A named individual should be responsible for anything AI touched, and that output should be checked before it leaves the building.
Judgment about fit. Decide when AI is an asset and when it's a crutch, because not every task needs automation. And a perceived strength is just a weakness.
This is now part of how I work with every client, not because I set out to work on AI, but because I can no longer do my job without addressing it. The businesses that put this foundation in place early will simply out‑decide the ones that don’t.